Characterisation of a cohort of opportunistically recruited patients with COVID-19 and approaches to patient stratification.

Multiomic biomarker discovery generally uses well curated cohorts of patients and health individuals, although this can present challenges in validation and deployment of biomarkers for diagnosis in “real world” situations. An alternative approach is to use cohorts of patients in routine clinical care, but this necessitates identification of relevant comparator groups. In infectious disease one option is to compare patients with a current infection with those who are convalescent. This approach has been applied using opportunistically collected biological samples from the Manchester Allergy, Respiratory and Thoracic Surgery Biobank COVID-19 cohort unimmunised hospitalised or convalescent patients for whom extensive clinical data and classical biomarker analysis are available. Pilot serum lipidomic and proteomic profiling was undertaken employing novel, rapid mass spectrometry methods. The cohort comprised 222 individuals, 25% of whom were of non-white ethnicity, 68% of whom were male and 66% were either overweight or obese. Around half (n = 116) were classified as severe based on symptomology using the WHO score, with a patient management score (the Manchester Severity Score) splitting the WHO “critical” category into two further sub-categories. Stratification of patients into those who had a current infection or were convalescent using both severity scores provided consistent results in analysis of classical cellular and biochemical markers. For example, C-reactive protein (CRP) was higher (mean 74.5 mg/L) in those with a current infection compared to those who were convalescent (mean 35.1 mg/L). Pilot multiomics analysis using novel methodology showed good reproducibility and identified lipid and protein biomarkers previously observed in COVID-19 infections including phosphatidyl cholines, triglycerides and C-reactive protein. The relative quantification of CRP by proteomics was correlated with conventional measurements. These data demonstrate the feasibility of using samples from patients who have either a current infection, or who are convalescent, as comparators and that the multiomics analysis pipeline is suitable for wider lipidomic and proteomic analysis in future.